It is estimated that the data volume between connected vehicles and edge/cloud server(s) will be about 100 petabytes per month by 2025. The networking framework we have, on the other hand, is the existing cellular network in which the most connected vehicles function today. However, such a network suffers from several issues and may not work under this predicted data demand. To address such a dilemma, a new paradigm, Vehicular Knowledge Networking (VKN), is recently introduced. In VKN, the data is transformed into knowledge and it is distributed with various lifetimes/relevance. To benefit from the knowledge, on the other hand, it should be placed intelligently such that a high number of vehicles can access and consume it. In this paper, we tackle this issue and propose mobility-aware smart knowledge placement. In the proposed method, vehicle mobility is analyzed to measure the centrality degree of a region. The computed centrality degrees are then further analyzed to identify the most central zones. The knowledge is placed on these zones to increase availability. We demonstrate the benefits of the proposed method through a simulation. Our preliminary result has shown that the mobility-aware smart knowledge placement makes knowledge accessible from vehicles over short range communication. Through such short-range availability of knowledge, vehicles can use the free spectrum to download it which decreases the cellular communication cost significantly.

Vehicular edge computing is enabling a variety of new services that better assist safety and comfort of driving. However, vehicles cannot offload an unlimited amount of computational tasks and input sensor data to a remote edge server because of the limitations in network bandwidth. In this paper, we design a learning-based task of floading mechanism that selects a small subset of input sensor data, which are expected to improve the application performance if processed by a rich and resource-intensive algorithm, hosted by the edge server. As a case study, we apply this framework to a vision-based object tracking application. The simulation results show that the proposed solution significantly improves object tracking accuracy with the same amount of resource consumption.

5G offers high flexibility at the radio, transport and core networks to support various services of critical verticals such as connected and automated driving. At the Radio Access Network (RAN), 5G defines a New Radio (NR). 5G NR utilizes different subcarrier spacing, slot durations, modulations and channel coding schemes. This flexibility offers the possibility to support automotive services with different and demanding requirements, such as Advanced Driver-Assistance System (ADAS), cooperative driving, and remote driving. Previous studies showed that 5G NR can be configured to achieve latencies below 2 ms. However, existing studies are generally restricted to scenarios with a limited number of users and unlimited bandwidth. Therefore, it is important to analyze whether 5G NR can effectively support these services as the network scales under limited spectrum allocations. This study advances the current state of the art to demonstrate that the capability of 5G NR RAN to support advanced V2X services depends on the RAN configuration (subcarrier spacing, slot duration and error protection) and network load.

Large-scale deployment of connected vehicles equipped with multiple automotive radar systems increases the demand on both the millimeter-wave (mmWave) automotive radar spectrum in 76–81 GHz and the vehicle-to-everything (V2X) communication spectrum in 5.9 GHz that is mainly allocated for the exchange of safety messages. To supplement V2X communication and support high data rates needed by broadband applications, the automotive radar spectrum with up to 4 GHz of contiguous bandwidth can be leveraged. For this purpose, various joint automotive radar-communication (JARC) systems have been proposed in the literature to perform both functions using the same radio-frequency (RF) signal and transceiver hardware. Combined with the high mobility in traffic, the directionality of RF transmission in mmWave spectrum and interference from other systems prevent JARC systems to achieve optimal communication and radar performance. In this work, we propose a dedicated neighbor discovery and medium access control (MAC) protocol for JARC systems to establish reliable communication links and improve the robustness of radar functionalities without requiring a separate control channel.

Vehicular micro cloud is a promising solution in which connected vehicles offer their resources as services to others and collaborate on tasks through vehicular networks. The properties of the vehicular micro cloud are generally set when it is formed at the core area and they remain static. However, a change in the environment could leave the micro cloud unable to respond adequately as its static properties become obsolete. Reacting to changes in the environment, on the other hand, requires dynamic updates on the properties of the vehicular micro cloud. In this paper, we address this dilemma and propose a chain of interdependent vehicular micro clouds. To respond to the changes in the environment, interdependent micro clouds are formed around the first vehicular micro cloud. Guidance information is transmitted before members of the dependent micro clouds arriving at the core area of the first micro cloud. Extensive simulations in different vehicular micro cloud settings show that the chain of interdependent vehicular micro clouds significantly improves the ability to respond to changes within the first micro cloud. Through the provided guidance information, members of dependent micro clouds are collectively prepared before they arrive at the core area.

Large-scale deployment of connected vehicles with cooperative awareness technologies increases the demand for vehicle-to-everything (V2X) communication spectrum in 5.9 GHz that is mainly allocated for the exchange of safety messages. To supplement V2X communication and support the high data rates needed by broadband applications, the millimeter-wave (mmWave) automotive radar spectrum at 76-81 GHz can be utilized. For this purpose, joint radar-communication systems have been proposed in the literature to perform both functions using the same waveform and hardware. While multiple-input and multiple-output (MIMO) communication with multiple users enables independent data streaming for high throughput, MIMO radar processing provides high-resolution imaging that is crucial for safety-critical systems. However, employing conventional precoding methods designed for communication generates directional beams that impair MIMO radar imaging and target tracking capabilities during data streaming. In this paper, we propose a MIMO joint automotive radar-communication (JARC) framework based on orthogonal frequency division multiplexing (OFDM) waveform. First, we show that the MIMO-OFDM preamble can be exploited for both MIMO radar processing and estimation of the communication channel. Then, we propose an optimal precoder design method that enables high accuracy target tracking while transmitting independent data streams to multiple receivers. The proposed methods provide high-resolution radar imaging and high throughput capabilities for MIMO JARC networks. Finally, we evaluate the efficacy of the proposed methods through numerical simulations.

Vehicles are expected to generate and consume an increasing amount of data, but how to perform risk reasoning over relevant data is still not yet solved. Location, time of day and driver behavior change the risk dynamically and make risk assessment challenging. This paper introduces a new paradigm, transferring information from raw sensed data to knowledge and explores the knowledge of risk reasoning through vehicular maneuver conflicts. In particular, we conduct a simulation study to analyze the driving data and extract the knowledge of risky road users and risky locations. We use knowledge to facilitate reduced volume and share it through a Vehicular Knowledge Network (VKN) for better traffic planning and safer driving.

Network virtualization and programmability allow operators to deploy a wide range of services over a common physical infrastructure and elastically allocate cloud and network resources according to changing requirements. While the elastic reconfiguration of virtual resources enables dynamically scaling capacity in order to support service demands with minimal operational cost, reconfiguration operations make resources unavailable during a given time period and may incur additional cost. In this paper, we address the dynamic cloud network control problem under non-negligible reconfiguration delay and cost. We show that while the capacity region remains unchanged regardless of the reconfiguration delay/cost values, a reconfiguration-agnostic policy may fail to guarantee throughput-optimality and minimum cost under nonzero reconfiguration delay/cost. We then present an adaptive dynamic cloud network control policy that allows network nodes to make local flow scheduling and resource allocation decisions while controlling the frequency of reconfiguration in order to support any input rate in the capacity region and achieve arbitrarily close to minimum cost for any finite reconfiguration delay/cost values.

This paper considers centralized routing over a network in which the cost of each edge is modeled as a random variable and the objective is to route the packets in a manner that minimizes the expected cost while constraining the variance of the cost to a pre- determined upper bound. This problem arises in the context of flow-based routing for wireless Software-Defined Networks (SDN). A Randomized Dijkstra-based Algorithm (RDBR) with a Lagrangian Relaxation Cost is proposed in which two distinct shortest paths are utilized randomly with an optimized randomization factor. Here the notion of shortest path identified by Dijkstra's algorithm refers to the property that these two paths both are of minimum regularized cost as defined by the expected total cost plus weighted variance. We prove the existence and optimality when the weight coincides with the optimal Lagrange multiplier and propose a sub-gradient descent method to compute the optimal multiplier. Numerical comparisons against other previously proposed solutions in the literature illustrate the performance improvements under RDBR at a significantly lower or comparable computational complexity.

Optical switches have been drawing attention due to their large data bandwidth and low power consumption. However, scheduling policies need to account for the schedule reconfiguration delay of optical switches to achieve good performance. The Adaptive MaxWeight policy achieves optimal throughput for switches with nonzero reconfiguration delay, and has been shown in simulation to have good delay performance. In this paper, we analyze the queue length behavior of a switch with nonzero reconfiguration delay operating under the Adaptive MaxWeight. We first show that the Adaptive MaxWeight policy exhibits a weak state space collapse behavior in steady-state, which could be viewed as an inheritance of the MaxWeight policy in a switch with zero reconfiguration delay. We then use the weak state space collapse result to obtain a steady state delay bound under the Adaptive MaxWeight algorithm in heavy traffic by applying a recently developed drift technique. The resulting delay bound is dependent on the expected schedule duration. We then derive the relation between the expected schedule duration and the steady state queue length through drift analysis, and obtain asymptotically tight queue length bounds in the heavy traffic regime.

All-optical switching networks have been considered a promising candidate for the next generation data center networks thanks to its scalability in data bandwidth and power efficiency. However, the bufferless nature and the nonzero reconfiguration delay of optical switches remain great challenges in deploying all-optical networks. This paper considers the end-to-end scheduling for all-optical data center networks with no in-network buffer and nonzero reconfiguration delay. A framework is proposed to deal with the nonzero reconfiguration delay. The proposed approach constructs an adaptive variant of any given scheduling policy. It is shown that if a scheduling policy guarantees its schedules to have schedule weights close to the MaxWeight schedule (and thus is throughput optimal in the zero reconfiguration regime), then the throughput optimality is inherited by its adaptive variant (in any nonzero reconfiguration delay regime). As a corollary, a class of adaptive variants of the well-known MaxWeight policy is shown to achieve throughput optimality without prior knowledge of the traffic load. Furthermore, through numerical simulations, the simplest such policy, namely, the Adaptive MaxWeight, is shown to exhibit better delay performance than all prior work.

This paper considers the end-to-end scheduling for all-optical data center networks with zero in-network buffer and non-negligible reconfiguration delay. It is known that in the regime where the scheduling reconfiguration delay is non-negligible, the rate of schedule reconfiguration should be limited in such a way as to minimize the impact of reduced duty-cycles and to ensure bounded delay. However, when the scheduling rate is restricted, the existing literature also tends to restrict the rate of monitoring and decision processes. We first present a framework for scheduling with reconfiguration delay that decouples the rate of scheduling from the rate of monitoring. Under this framework, we then present two scheduling algorithms for switches with reconfiguration delay, both based on the well-known MaxWeight scheduling policy. The first one is the Periodic MaxWeight (PMW), which is simpler in computation, but requires prior knowledge of traffic load. The other is the Adaptive MaxWeight (AMW), which, in contrast, requires no prior knowledge. We show the stability condition for both algorithms and evaluate their delay performance through simulations.

A novel architecture for the future data center networks with possibly up to a thousand of Top of the Rack (ToR) switches is proposed. The proposed architecture, WiCOD, relies on a wireless control plane serving an all-optical data plane. The first contribution of the work is the separation of the data and the control planes: while the data is switched between the ToR switches in an all-optical high rate network, the network state and control information is continuously conveyed to and from a central unit over an ultra low-latency wireless network. A proof of concept for this architecture is also presented by considering the initial design possibilities for each one of the planes. In order to obtain low packet delays, the data plane scheduling policies take non-zero reconfiguration and monitoring delays into consideration. The results prove that very low queueing delays are guaranteed for strictly frequent updates on the network state. Based on this observation, a technique for monitoring of ToR switch queue occupancy information is purposed. This monitoring technique uses mmWave wireless communications via a spatially adapted MIMO Orthogonal Frequency Division Multiple Access (MIMO-OFDMA) over a static frequency selective channel with large number of densely packed ToRs. The reduced monitoring delays, offered by this low-latency radio access technology, makes the fine-grained and adaptive circuit switching feasible and, in turn, enables a high utilization of optical switches.

In-network queuing in the Internet-style networks enables the distributed operation and scalability across the network at the cost of excessive delay and tardy flow completion times. Data center networking, in contrast, are proposed to depart from this classical approach and avoid in-network queuing all together. In this new class of network solutions serving interdata center traffic, a densely packed fairly local area network of stationary end hosts are often managed by a single entity, allowing for fine-grained management and scheduling of flows across the data center. The overall objective of this work is to develop a framework, from first principles, which relies on the unique attributes of data centers to propose a transformative novel networking architecture with increased level of efficiency and significantly smaller latency. By separating the control and data planes, the proposed hybrid architecture avoids in-network queuing and results in significantly lower delay. The critical technical challenge is to design endend circuit switching mechanisms that account for monitoring as well as circuit reconfiguration delays. Furthermore, the design has to minimize the computational complexity of the scheduling algorithm as well as the cost of monitoring across the network. In this context, this paper underlines a family of recent technologies and networking advances as promising enablers and discusses the most significant set of challenges.

The Internet of Vehicle (IoV) network refers to networking technologies that could support highly economical and effective communication channels for interactions between vehicles. The success of IoV relies on efficient vehicle-to-infrastructure (V2I) or vehicle-to-vehicle (V2V) communications. Among all existing communication standards, the IEEE 802.15.4 standard, which comprises a simple physical (PHY) layer and an energy-efficient medium access control, is a promising candidate for IoV networks. In this paper, we propose analytical models for IEEE 802.15.4 nonbeacon-enabled mode for IoV by considering two major features, i.e., non-saturated traffic pattern and large-scale network, in IoV applications. We investigate the performance of IEEE 802.15.4 nonbeacon-enabled mode for IoV. Our study can provide guidelines for vehicles to dynamically adjust the broadcasting rate and achieve a higher probability of success in such communications.

This paper presents a novel learning-based method for single image super-resolution (SR). Given an input low-resolution image and its image pyramid, we propose to perform context-constrained image segmentation and construct an image segment dataset with different context categories. By learning context-specific image sparse representation, our method aims to model the relationship between the interpolated image patches and their ground truth pixel values from different context categories via support vector regression (SVR). To synthesize the final SR output, we upsample the input image by bicubic interpolation, followed by the refinement of each image patch using the SVR model learned from the associated context category. Unlike prior learning-based SR methods, our approach does not require the reoccurrence of similar image patches (within or across image scales), and we do not need to collect training low and high-resolution image data in advance either. Empirical results show that our proposed method is quantitatively and qualitatively more effective than existing interpolation or learning-based SR approaches.

We propose a novel learning-based method for single image super-resolution (SR). Given a low-resolution input image and its image pyramid, we advance a context-constrained image segmentation to construct a super-pixel database with different context categories for learning purposes. By utilizing context-specific image sparse representation, our method aims at modeling the relationship between the interpolated image patches and their ground truth pixels from different context categories via support vector regression (SVR). To produce the final SR output, we upsample the low-resolution input, followed by the refinement of each image patch using the SVR models observed from the associated context categories. Unlike prior learning-based SR methods, our approach advances a self-learning technique and does not assume the reoccurrence of image patches (within or across image scales). We do not need to collect training low/high-resolution image data in advance either. Empirical results verify the effectiveness of our SR approach, which quantitatively and qualitatively outperforms existing interpolation or learning-based SR methods in most cases.